This page contains a Flash digital edition of a book.


Keeping an eye on batteries

There are six 2V cells in every 12V battery and 40 batteries in the string that provides the backup power for a typical UPS. If one of those 240 cells fails to an open circuit state, you have zero volts just when you really need them. By Phil Thane

Fortunately for data center operators, conventional lead-acid batteries and their modern incarnation - the valve regulated type (VRLA) - are very reliable but when any one of 240 components in a string can bring down the whole system, it takes a lot of trust to rely on inherent reliability, especially if it is intangible. With this in mind it is unsurprising that more and more data centers are improving their battery monitoring techniques.

DEALING WITH REDUNDANCY There are three ways of dealing with the potential for battery failure in data centers (and any other operation that needs an assured supply). Redundancy, planned maintenance and monitoring.

Redundancy in a system with 240 single points of failure isn't a cheap option, you can't switch out a failed cell or failed battery quick enough when the utility company supply fails; to have an

Batteries inside cabinets. Image courtesy of APC

N+1 system you need a complete set of batteries in parallel with the first, and a second UPS. Then how are you going to calculate the risk of a failure somewhere in that second, parallel string?

Many data centers rely on planned maintenance and asset management. Planned maintenance usually takes the form of manually testing each battery every quarter and replacing any whose performance is significantly below the rest. The asset management part is simply talking to the supplier about the batteries, the operating conditions, how often the UPS has called on

them, then taking an educated guess on when is good time to replace them all.

Batteries can fail suddenly, even one day after a quarterly test you cannot be sure they are all working properly. Increasing the testing frequency might help a little but it's a time consuming job (therefore expensive) and the testing current fed into the batteries to carry out the tests can actually have the effect of reducing their life. Asset management isn't optimal either; in order to reduce the likelihood of failure, most operations will err on the side of caution and replace costly batteries that could have remained in place a lot longer.


The IEEE 1491-2005, “Guide for Selection and Use of Battery Monitoring Equipment in Stationary Applications” was released at the end of 2005. It was the result of nearly 10 years of discussion between all the interested parties who had to pare this down to the most important seventeen areas for consideration.

The working group committee is currently working on a revision, expected to be released later this year which will define a total of 25 battery performance parameters to monitor. At present no device on the market measures all 17 in the original list, they concentrate on those most germane to predicting battery performance and life.

PREDICTING FAILURE Stationary fixed battery monitoring systems carry out much more frequent checking, often checking more parameters than a technician with a hand-held monitor. Crucially they generate enough data to show trends in performance enabling data center staff to predict when a battery will fail and arrange to replace it in time. Using data collected on an entire string of batteries over an extended period, managers are able to make a rational decision about asset management, leaving a


Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20